Eur Respir J 1989, 2, 522-527

Exercise tolerance in chronic obstructive pulmonary disease: importance of active and passive components

of the ventilatory system

A. Loiseau *, C. Dubreuil**, P. Loiseau*, J.C. Pujet**, A. Georges*, G. Saumon*

Exercise tolerance in chronic obstructive pulmonary disease: importance of active and passive components of the ventilatory system."A. Loiseau, C. Dubreuil, P. Loiseau, J.C. Pujet, R. Georges, G. Saumon. ABSTRACT: We investigated which components or ventilatory function are related to exercise tolerance In cbronlc obstructive pulmonary disease (COPD) patients. Physical characteristics, usual lung function, tlmlng and neuromuscular components of ventilation were measured In 113 out­patie.nts in whom FEV ;vc was less than 75% of the predicted value and exercise was limited by breathlessness. These variables were used to pre­dict the maximum work load during progressive bicycle e.xercise. The prediction was obtained using a stepwise procedure in men and women separately. Among the variables selected, age, body weight, FEV/VC, l'Imax ' and P0.1NTffi accounted for 79 % llf the variability in maximum performance in men. The predictive model was statistically verified and was stable. The mean prediction error was 12 Watts. Among these variables, P0./VTffi, Pr •. and FEV/VC were the main determinants of maximum work load (MWL). These results show that exercise limitation in COPD is related to impairment of both the active (inspiratory muscles) and passive (respiratory impedance) components of the ventilatory system. The same conclusions concerning passive components are proposed for women, despite a smaller population which prevented ver111catlon of the prediction. Eur Respir 1., 1989, 2, 522-527.

• INSERM U.82; Faculte Xavier Bichat, 16 Rue Henri Huchard 750 18 PARIS, France. - Centre de Traitement des Affections Respiratoires: 74 Rue de la Colonic 750 13 PARIS, France.

Correspondence: A. Loiseau, INSERM U.82, Faculte Xavier Bichat, 16 rue Henri Huchard 750 18 PARIS, France.

Keywords: COPD; exercise; impedance; maximum work load; prediction; stepwise regression analysis.

Received: March 1988; accepted for publication: January 13, 1989.

Maximal exercise performance has proved useful for assessment of respiratory impairment in patients with chronic obstructive pulmonary disease (COPD) [1). Exercise hyperpnoea is the consequence of the interac­tion of chemical, neural, muscular, haemodynamic, and mechanical processes [2]. Exercise tolerance in COPD patients is only explained in part by usual assessment of ventilatory limitation (FEV

1NC), static lung volumes, or

TLCo [1, 3-6]. Pulmonary haemodynamics do not appear important in limiting exercise in COPD patients [7, 8]. Other factors must therefore interfere. We hypothesize that neuromuscular drive and respiratory efficiency might be among these factors. For instance, maximum exercise may depend on the ability of inspiratory muscles to sustain the required ventilation. The importance of respiratory muscle fatigue as a component of exercise limitation has been demonstrated [9]. Recent studies have documented central neural failure during high ventilation and have acknowledged its role in ventilatory failure (10-12). Based on this assumption one could expect that the maximum exercise performance could be more precisely predicted by taking into consideration additional factors such as occlusion pressure (P0), respiratory timing (TI!fToT). inspiratory flow rate (Vr{TJ) [13], maximum inspiratory pressure (Pim .. ) [14], and respiratory impedance (P0. 1Nr/ TI) [15, 16).

We investigated which components of ventilatory function are related to exercise tolerance in COPD pa· tients. We selected a set of predicting variables measured at rest that might determine exercise performance, in­cluding physical characteristics, lung volumes, FEV

1, and

estimates of respiratory drive. The respective imponance of these factors in predicting exercise limitation in COPD patients was assessed using stepwise and ridge regres­sions. We found that respiratory muscle strength and respiratory impedance are important determinants of exercise tolerance in these patients.

Methods

Patients population

We selected 113 COPD out-patients (93 men, 20 women) with clinical and physiological evidence of chronic airflow limitation according to the criteria of the American Thoracic Society [17]. Twenty two also pre­sented emphysema and 11 post-tuberculosis fibrotic lesions. Patients with acute bronchial infection, cardio­vascular disease or history of asthma were excluded. None of these patients had received any drug during the two weeks prior to the study. In all patients, FEVJVC was

EXERCISE PERFORMANCE IN COPD 523

less than 75% of the predicted value and exercise was limited by breathlessness. 80% of patients presented objective evidence of exercise limitation of respiratory origin: maximal cardiac frequency during exercise was less than 80% of the predicted maximal value, maximal exercise ventilation was in keeping with FEV

1 [3] . Several

factors might have influenced exercise limitation in others (such as cardiovascular factors, leg muscle fatigue, but also ventilatory limitation). Most of the subjects were smokers and 69% were still smoking an average of one pack of cigarettes per day. All patients were familiar with pulmonary function testing and gave informed consent for this study. The physical characteristics and lung function tests of the population are given in table 1. The predicted values for lung function tests are those of QUANJER et al. (18).

Table 1. - Physical and functional values in COPD patients

Men (n=93) mean S BM

Age yrs 62 1.0 Height cm 169 0.8 Weight kg 67 1.2 VC %prcd• 62 2.6 FEY/VC % 39 1.2 RV!I'LC % 54 0.9 Pao2 kPa 9.2 0.16 Paco

2 kPa 5.3 0.07 Sao2

% 93.5 0.4 Pi

max kPa 4.8 0.17

PO.lNT(fJ kPa·t1·s 0.41 0.017

*: percent of predicted value.

Methods

Static lung volumes were measured with a water-sealed spirometer (Spirotest 3, Jaeger, West Germany); residual volume was measured by helium dilution. Blood gases were determined from an arterialized ear lobe blood sample [19] at rest (IL213, Delhomme, Paris). An auto­matic apparatus was used to measure mouth occlusion pressure (P0) [20]. Inspiratory resistance of the appara­tus was 0.15 kPa·l-1·s and the expiratory resistance 0.05 kPa·Fs. Maximal static inspiratory pressure (Pim .. ) was measured at FRC during total occlusion with a ±15 kPa pressure transducer (MP45, Validyne, Northridge, CA). Tidal volume (VT), inspiratory (TI) and expiratory (TE) times from the breath preceding occlusion were meas­ured with a Fleish no. 3 pneumotachometer and a ±0.2 kPa pressure transducer (Validyne). Calibrations were made by the usual techniques: a static calibration for P

0.1

and Prm .. transducers, and a dynamic calibration with a calibrated syringe for the pneumotachometer [21]. Exer­cise was performed using a cycle ergometer that dissi­pates the work of pedalling against an electromagnetic brake (Gauthier EPC 7701, Paris, France). It was cali­brated by means of a standard weight attached to a pedal

which provided a known torque against the electromag­netic brake.

Exercise protocol

Progressive exercise was performed at a cycling speed of about 50 rpm by 102 patients, after a 3 min rest. The work load was increased by 30 watt steps every 3 min until exhaustion. In 11 male patients, the work load was increased by 10 watt steps every minute until exhaustion and the exercise was performed in triplicate in the same week to determine intra-subject variability. Breathing patterns at rest, P0. 1 and PI., .. were obtained during the resting period before exercising. They represented the mean of 6 to 10 non-consecutive breaths.

Women (n=20) (range) mean SBM (range)

(22-77) 63 2.2 (49-80) (151- 184) 162 0.1 (154-170) (40-91) 56 2.6 (37-87) (47- 100) 74 5.8 (29-100) (20-71) 48 3.7 (19- 83) (33- 68) 56 1.7 (42-72)

(7.1- 11.6) 9.6 0.21 (8.4-10.7) (4.0--6.9) 5.2 0.12 (4.3-5.9) (86-96) 94.5 0.4 (92-96) (2.0--8.3) 4.2 0.41 (2.2-9.0)

(0.09- 1.05) 0.54 0.068 (0.22- 1.34J

Statistical analyses

Data are presented as mean±sEM unless indicated oth­erwise. The relation between variables was obtained by linear regression using a least square method. Stepwise multiple regression, and ridge procedure (22, 23, 24) were used to determine a predictive equation for maxi­mum performance in men and women separately. Maxi­mum performance was defined as the largest work load performed allowing longer exercise than half a step duration. The analyses were performed using STATGRA­PHICS software (STSC, Rockville, Maryland, USA) on an IBM PC computer.

Stepwise regression consists of selecting those vari­ables which provide the best information for the pre­diction using the partial correlation coefficients as a measure of the predictor importance of variables. The coefficient of determination (R2) represents the overall fraction of variability for the predicted variable that is explained by the regression model. However, when many predictor variables are interdependent, the regression coefficients may not be stable from one population sample to another and cannot be used in a predictive mode. Thus, a further analysis was performed. Ridge regression [23]

524 A. LOISEAU ET AL.

was used to verify the stability of regression coefficients. The reliability of the predictive equation was tested using another sample of the patient population.

The relative importance of each of the components in the predictive equation was approximately determined from the standard partial regression coefficient values [24].

The 95% confidence interval (±2 so) for the predicted­observed MWL relationship was used as a criterion of the soundness of classification; the patients within the confidence interval were considered as correctly classi­fied. The confidence interval was obtained from the mean of the intra-subject MWL standard deviation.

The COPD population was divided into 2 groups according to sex (group A: 20 women, group B: 93 men). Intra-subject variability of MWL was assessed in 11 subjects in group B (subgroup B3). The other subjects in group B (n=82) were randomly allocated to 2 sample groups of equal size and identical MWL distribution (subgroup B1 and B2). Group Bl (n=41) was used to establish the prediction and groups B2 and B3 (n=52) were used to verify the prediction model.

"Predictor" variable set

Age, height and body weight were chosen because we felt that these factors account for the variability in the subject's physical aptitude for exercise. VC(% predicted) and RV(ILC were selected because they are indexes of ventilatory capacity and thorax shape while FEV1NC reflects the resistive load of the ventilatory system dur­ing expiration. Pao

2, Sao

2• and Paco

2 were selected as

indexes of hypoxaemia and capnia. Mean inspiratory flow (VT/TI) and the inspiratory duty cycle (Tr!fToT) were chosen as components of ventilation depending on neu-

Table 2. - Stepwise regression

a) In men: group Bl (n=41)

Y-intercept=l4W (Multiple) R2==0.786

Regression coefficient so

b) In women: group A (n=20)

Y -intercept=172W

Regression coefficient so

Prmax kPa

5.81 1.56

p<O.OOl

R~0.798

Age

yrs

-1.60 0.341

p<O.OOl

Age

yrs

-0.60 0.295

p<0.05

Variable

PO.lNT!fl kPa·t1·s

-75.56 10.95

p<O.OOl

raJ events. The inspiratory effectiveness of muscle con­traction was taken into account by means of PI and, passive inspiratory mechanical impedance of the ~entila­tory system by means of the effective inspiratory impedance (P0.1NT/TI). P0•1 was chosen as an index of inspiratory neuromuscular drive.

Results

Physical characteristics and the results of lung func­tion testing in COPD patients are given in table 1. Pa­tients had decreased vital capacity (64±2.4% predicted value), airway obstruction (FEV1NC=41±1.2%) and increased RV(fLC ratio (54±1.0%). Analyses of blood gases revealed hypoxaemia (Pao

2=9.3±0.14 kpa), de­

creased haemoglobin saturation (Sao2=93.7±0.34%) and normocapnia (Paco

2=5.3±0.05 kpa). The MWL accord­

ing to our definition ranged from 0 to 120 Watts. Only 2 men and 2 women were unable to perform on the lowest work load for less than 1 min 30 (0 Watt). The results of lung function were not significantly different between the B 1 and B2 groups. Significant correlations were observed between MWL and age, FEV

1NC, RV/

lLC, P0.1NT/TI, and Pr., .. (p<.OOI).

Stepwise procedure

The stepwise multiple regression selected age, FEY/ VC, weight, Pim .. and P0•1NT/TI from the "predictor" variable set for men. Values and levels of significance of the regression coefficients are given in table 2. 79% of the variability in performance was explained by this model, however, this figure might be due in part to some interdependence between the predictor variables.

Variable

P0_/VT(fl

kPa·l"1·s

-83.7 13.67

p<O.OOl

FEV1NC

%

0.525 0.20

p<0.02

FEV1NC Weight

% kg

1.39 0.763 0.232 0.244

p<O.OOl p<0.004

Predicted variable: maximal work load (Watt); "Predictor" variable set: age, height, weight, VC% pred, FEV.f VC, RV/fLC, VT(fi, Tt!fTOT' Pao2, Paco2, Sao2, P0.l' Pimax' P0.1NT!fr

EXERCISE PERFORMANCE IN COPD 525

Correlation analysis showed significant interdependence between FEV/VC and age only (p<O.OS) and a border­line interdependence between PI., •• and P0_./VT{fi (p<0.06). The ridge trace did not reveal instability in the coefficients. P0.1NT!fi, FEV,NC and Pim .. were the most important factors in the prediction equation.

In women, the same amount of MWL variability was explained by age, FEV,NC and P0.1NT{TI (80%). Val­ues and levels of significance for the regression coeffi­cients are given in table 2. However, the small size of the population sample for women prevented in depth analy­sis.

The intra-subject variability (SD) of MWL was inde­pendent of the MWL level. The mean of the intra-subject standard deviations was equal to 6 Watts. As it was far smaller than the work load corresponding to a single step, we refrained from testing the reproducibility of the 30 W /3 m in protocol.

150 / /

/ /

0 / / / /

/~ / /

120 / / / /

/ . / /

I / /

·-/ / ~ / 0 ;,; /., /

/ 90 / /

..J . /.~ / ;,; ~ :// / r/ '0 .. . ~

/ /0 / • / • 0 /

'0 60 ;:;u ~ / Q. / V e/ / •

/

30 //•/o //

~ ~' 0

,//,:/

0 30 60 90 120 150

Observed MWL ( W I

Fig. 1. - Plot of maximum work load (MWL) observed during pro­gressive exercise in COPD patients vs MWL predicted according to the multiple linear regression equation. Oosed circles correspond to the 30 W/3 min and open c.ircles to the 10 W/1 min progressive protocols for groups B2 and 83 (52 patients). This population sample was used to verify the prediction equation established with the sample group Bl (41 patients) with similar MWL distribution.

Validity of the model

The prediction equation was verified using the samples B2 (n=41) and B3 (n=ll) (figure 1). Prediction error ranged from -38 W to +37 W (mean value=12 W). There was no major difference in the prediction error when the results of the two exercise protocols were compared (1 0 W or 30 W step load). There was also no significant difference in the prediction when comparisons were made using the predictive equation from the stepwise regres-

sion or the ridge coefficients. The percentage of patients who were correctly classified was 66% in sample B2 (figure 1; closed circles). The maximum work load was underestimated for less than a step (23 W) in 4 patients, and overestimated in 10 patients. Nevertheless, it should be emphasized that in 8 out of 10 of the latter, the pre­dicted value could be accurate, since the predicted maximum work load was less than the value for the next step in the protocol. For example, a subject with a pre­dicted maximum workload of 115 W and with an ob­served perfonnance of 90 W could actually be capable of performing 115 W. However, since the intermediate levels were not tested in these patients, the observed MWL may have been underestimated. Thus, the percentage of patients with good prediction lies between 66 and 85% (mean 75%). The reliability of the prediction equation was similar for the subjects in sample B3 (8 out of 11 correct predictions).

Discussion

The relationship between maximum exercise and pul­monary function measurements in COPD patients at rest was always found to be weak [1, 3-5]. The best predic­tion for MWL was obtained using the following relation­ship [1]: MWL (kgm·min-1)=563-5.4 Age+142 FEV1+7.5 Ttco; R2=0.54). This model was not verified with an­other group of patients. The present study shows that the inclusion of respiratory muscle strength and impedance of the ventilatory system as additional factors improves MWL prediction (R2=0.79).

Methodological considerations

Various prerequisites must be fulfilled to establish a valid predictive model: physiological significance of the retained variables as will be discussed, adequacy of the signs of the regression coefficients, accuracy of the prediction (R2), absence of interdependence between the retained variables (assessed by the stability of the regres­sion coefficients) and validation of the model using another sample of patients from the same population.

The algebraic sign ( + or -) of each multiple regression coefficient was found to be as expected according to the physiological effect of the variable on MWL (inverse or positive relation): it corresponds to a positive effect on MWL for Pim .. • FEV1NC and weight, and to a negative effect for age and P0.1NT!fl. Despite the interdepend­ence of FEV1NC and age, the stability of the regression coefficients was correct as assessed by the ridge trace. Verification of the MWL prediction in samples B2 and B3 confirmed the stability of the model since the MWL of about 75% of patients were correctly predicted. The accuracy of the prediction (12 watts) is acceptable since variability is expected to be large in COPD patients weakened and stressed by their disease.

Physiological considerations

The components of the MWL prediction equation are more or less directly related to three main determinants

526 A. LOISEAU ET AL.

of exercise aptitude that is, the physical ability to per­form exercise and particularly muscle mass, the passive respiratory system, the active respiratory system. a. Physical abili ty is accounted for in the prediction equation by age and body weight which are regularly found to be predictors of Vo2max, which in turn, is directly related to muscular ability in healthy non obese subjects [25]. The observation that age affects MWL in normal subjects, was also observed in this series of COPD patients (despite large variations in disease progression). Age probably reflected leg muscular weakness and ar­ticular stiffness in these sedentary patients. Body weight, in non obese COPD subjects, is indicative of muscle mass [25], an important factor in exercise capacity. These patients, and particularly those with emphysema, often have muscle atrophy because of nutritional depletion [26). b. Airway obstruction is accounted for in the regression by FEY/VC which indirectly quantifies the increase in expiratory impedance. In COPD patients, maximum exercise ventilation usually approaches, or may even exceed, maximal voluntary ventilation [3, 27] and is limited by the maximal expiratory flow rate [28). During incremental exercise testing, the need for increased air­flow soon forces the subject to reach the maximum value obtained during a forced volitional effort for a given lung volume [29, 30]. This mechanical limitation of maximum exercise ventilation in COPD patients is the consequence of both airway obstruction and of hyperin­flation. c. The efficiency of the active respiratory system de­pends on both the inspiratory neuromuscular drive and the impedance which opposes inspiration. P

0.1NT{n may

be considered to be an index of the effective inspiratory impedance [15, 16] and Prm .. as a measure of inspiratory muscle strength [14] . . Both parameters are related to the efficiency of the active ventilatory system and also tho­rax shape. Pr,. .. appears to be a valuable parameter to quantifying loss of respiratory muscle efficiency as a consequence of hyperinflation. During hyperinflation, the inspiratory muscles are inCI less efficient force-length relationship due to a shorter operating length. As the diaphragm flattens, greater activity of the inspiratory intercostal-accessory muscles is required to provide an increase in ventilation [31 ].

The factors retained in the prediction of MWL are therefore directly or indirectly related to systemic or respiratory muscles and to the impedance of the ventila­tory system. These findings are consistent with those of Kn.uAN and coworkers [32, 33) who demonstrated that these same factors are involved in the development of exercise dyspnoea. In conclusion, these factors therefore appear to be of importance in determining maximum performance and should be taken into consideration when searching for therapies (nutrition, muscle contractility, dyspnoea sensation ... ) aiming at increasing exercise tol­erance in COPD patients.

Other factors to be considered

Our prediction of MWL is still imperfect. It indicates that other factors need to be considered. The stress re-

suiting from a relatively strenuous exercise and leg weakness may play a role in interrupting exercise. These factors are not easy to quantify due to the absence of corresponding indexes at rest. Dyspnoea tolerance might also be a factor to take into consideration because it would account for the subjective components of effort tolerance. The model would perhaps be improved by the addition of TLCO. TLco reflects the effectiveness of gas exchange and thus might be a valuable predictor of how the respiratory system allows the patient to sustain effort. It has been found to be correlated with exercise tolerance [1, 51 and is probably one of the factors that could improve the prediction from measurements performed at rest. Al­though cardiovascular factors do not seem to affect maximum exercise to a large extent in COPD patients [7, 8], they might provide additional information explaining exercise limitation.

Arterial blood gases were not retained in stepwise regression for evaluation. This confirms the previous observations that blood gases are not closely related to exercise performance [34]. Metabolic acidosis does not usually occur during maximum exercise in COPD. The maximum blood lactate concentration remains in the normal resting range [35, 36]. Despite hypoxaemia, oxygen delivery to exercising muscle is usually main­tained [1].

The multiple linear regression model is limited to additive effects of predictor variables, whereas interac­tive effects are perhaps more plausible on physiological grounds. Integration of interactive effects in an explica­tive model would therefore be advantageous. However, this implies that all possible relations between the vari­ables studied are known. This problem however is far from being solved.

In conclusion, this study shows that maximum exer­cise performance in COPD patients is related to both active and passive components of the ventilatory system and probably depends on them. The clinical relevance of this finding is that therapies aiming at increasing respi­ratory muscle strength and decreasing respiratory imped­ance might be useful for submaximal exercise tolerance in these patients.

Acknowledgement: We wish to thank Ms. F. Miklovic for typing the manuscript

References

1. Jones NL. Jones G, Edwards RHT. - Exercise tolerance in chronic airway obstruction. Am Re·v· Respir Dis, 1971, 103, 477-491. 2. Dempsey JA, Vidruk EH, Mitchell GS. - Pulmonary control systems in exercise: update. Fed Proc, 1985, 44, 2260-2270. 3. Cotes JE. Posner V, Reed JW.- Estimation of maximal exercise ventilation and 0, uptake in patients with chronic lung disease. Bull Europ Physiopath Resp, 1982, Suppl. 18, 221- 228. 4. Gilbert R, Keighley J, Auchincloss JH (Jr). - Disability in patients with obslructive pulmonary disease. Am Rev Respir Dis. 1964, 90, 383-394. 5. Cotes JE, Zedja J and King B. -Lung function impair­ment as a guide to exercise limitation in work-related lung disorders. Am Rev Respir Dis, 1988, 137, 1089-1093.

EXERCISE PERFORMANCE IN COPD 527

6. Brown SE, IGng RR, Temerlin SM, Stanbury DW, Mahutte K, Light RW. - Exercise performance with added dead space in chronic airflow obstruction. J Appl Physiol: Respiral Envi­ron Exercise Physiol, 1984, 56, 1020-1026. 7 . Adnot S, Andrivet P, Piquet J, Brun-Buisson C, Rauss A, Bignon J, Lemaire F. - The effects of urapidil therapy on a haemodynamics and gas exchange in exercising patients with chronic obstructive pulmonary disease and pulmonary hypertension. Am Rev Respir Dis, 1988, 137, 1068- 1074. 8. Dal Nog are AR, Rubin U. - The effects of hydralazine on exercise capacity in pulmonary hypertension secondary to chronic obstructive pulmonary disease. Am Rev Respir Dis, 1986, 133, 385-389. 9. Roussos CS, Fixley MS, Gross D, Macklem PT. - Respi­ratory muscle fatigue in man at FRC and higher lung volumes. Physiologist, 1976, 19, 345-349. 10. Watchko JF, Standaert TA, Mayock DE, Twiggs G, Woodrum DE. - Ventilatory failure during loaded breathing: the role of central neural drive. J Appl Physiol: Respirat Environ Exercise Physiol, 1988, 65, 249-255. 11. Bazzy AR, Haddad GG. - Diaphragmatic fatigue in unan­aesthetized adult sheep. J Appl Physiol: Respirat Environ Exercise Physiol, 1984, 57, 182-190. 12. Bellemare F, Bigland-Ritchie B. - Central components of diaphragmatic fatigue assessed by phrenic nerve stimulation. J Appl Physiol: Environ Exercise Physiol, 1987, 62, 1307- 1316. 13. Milic-Emili J, Grunstein MM. - Drive and timing compo­nents of ventilation. Chest, 1976, 70, suppl., 131- 133. 14. Criee CP, Neuhaus KL, Whilhems E, IV-euzer H. - Mouth occlusion pressure in patients with chronic obstructive lung disease in stable state and in acute respiratory failure. Bull Europ Physiopath Resp, 1982, 18, 155-164. 15. Derenne JP, Couture J, Iscoe S, Whitelaw WA, Milic­Ernili J. - Occlusion pressures in men rebreathing eo. under methoxyflurane anesthesia. J App/ Physiol: Respirat Environ Exercise Physiol, 1976, 40, 805-814. 16. Milic-Emili J. - Recent advances in the evaluation of respiratory drive. lnt Anesthesiol Clin, 1977, 15, 39- 58. 17. ATS. - Chronic bronchitis, asthma and pulmonary em­physema: a statement by the committee on diagnostic standards for non tuberculosis disease: respiratory disease definitions and classifications. Am Rev Respir Dis, 1962, 85, 762- 768. 18. Quanjer PH, ed. - Standardized lung function testing. European Community for coal and steel. Bull Europ Physiopalh Resp, 1983, 19, 45- 51. 19. Godfrey S, Wozniack ER, Courtenay ERJ, Samuels CS. ­Ear lobe blood samples for blood gas a:talysis at rest and during exercise. Brit J Dis Chest, 1971, 65, 58- 64. 20. Delavault E, Saumon G. - An analog device to facilitate occlusion pressure measurements. Pjlugers Arch, 1980, 385, 259-261. 21. Varene P, Vieillefond H, Saumon G, Lafosse JE. - Eta­lormage des pneumotachographes par m6thode integrale. Bull Europ Physiopath Resp, 1974, 10, 349- 360. 22. Ralston A, Wilth HS. -Mathematical methods for digital computers. New York. J. Wiley andsons,1964, vol.l, 196--199. 23. Draper N, Smith H. - Applied regression analysis, second edition. New York, J. Wiley and sons, 1981, 313- 325. 24. Snedecor and Cochran. - Statistical methods. Ames, Iowa State University Press, 1967, 381-418. 25. Wasserman K, Hansen JE, Sue DY, Whipp BJ. - Prin-

ciples of exercise testing and interpretation. Philadelphia, Lea and Pebiger, 1987, 72- 75. 26. Openbrier DR, Irwin NM, Rogers RM, Gottlieb GP, Dauber JH, Van Thiel DH, Pennock BE.- Nutritional status and lung function in patients with emphysema and chronic bronchitis. Chest, 1983, 1, 17-22. 27. Clark TJH, Freedman S, Campbell EJM, Winn RR. -The ventilatory capacity of patients with chronic airway obstruction. Clin Sci, 1969, 36, 307- 319. 28. Ogilvie CM, Stone RW, Marshell R. -The mechanics of breathing during the maximum breathing capacity test. Clin Sci, 1955, 14, 101- 107. 29. Leaver DJ, Pride NB. - Flow-volume curves and expira­tory pressures during exercise in patients with chronic airway obstruction. Scand J Respir Dis, 1971, Supp1. 77, 23- 27. 30. Stubbing DG, Pengelly LD, Morse JLC, Jones NL. -Pulmonary mechanics during exercise in subjects with chronic airflow obstruction. J Appl Physiol: Respirat Environ Exercise Physiol, 1980, 49, 511- 515. 31. Dodd DS, Brancatisano T, Engel LA. - Chest wall me­chanics during exercise in patients with severe chronic airflow limitation. Am Rev Respir Dis, 1984, 129, 33- 38. 32. Killian KJ, Campbcll EJM. - Dyspnea and Exercise. Ann Rev Physio/, 1983, 45, 465-479. 33. Killian KJ, Jones KL. - The use of exercise testing and other methods in the investigation of dyspnea. Clin Chest Med, 1984, 5, 99- 108. 34. Jones NL, Berman LB. - Gas exchange in chronic airflow obstruction. Am Rev Respir Dis, 1984, Suppl. 129, S81- 83. 35. Kanarek D, Kaplan D. Kazerni H. - The anaerobic thresh­old in severe chronic obstructive lung disease. Bull Eur Physio­path Resp, 1979, 15, 163- 169. 36. Saunier C, Gimenez M, Harteman D. - Blood glycosis during muscular exercise in patients with chronic lung disease and hypoxemia. Bull Eur Physiopath Resp, 1979, 15, 291- 302.

A. Loiseau, C. Dubreuil, P. Loiseau, J .C. Pujet, R. Georges, G. Saumon. RESUME: Chez 113 malades ambulatoires souffrant de BPCO, nous avons recherche les composantes de la fonction ventila­toire qui influencent la performance maximale a !'effort (MWL) durant un exercice progressif. Des parametres d 'aptitude phy­sique et de fonction pulmonaire ainsi que les composants actifs et passifs de la ventilation ont ete selectionnes comme predicteurs possibles de MWL. Chez tous ces patients, le VEMS/ CV etait infcrieur a 75% de sa valeur theorique et l'exercice etait limite par une dyspnee d'effort. La prediction de MWL a ete effectuee s6parcment chez les hommes et les femmes, en utilisant une regression multiple pas a pas. La variabilite de la MWL est expliquee pour 79% par !'age, le poids, le VEMS/ CV, la PI,..., et P~JVT{fr chez les hommes. La verification statistique du modele a montre que la prediction de MWL etait stable avec une erreur de prediction de 12 Watts. P~,NT!fl, la PI ... et le VEMS/CV etaient les determinants majeurs de la MWL. Ces resultats montrent que la limitation ventilatoire a !'exercise chcz les BPCO est en partie determinee par les composants actifs (force musculaire inspiratoire) et passifs (impedance) du systeme ventilatoire. La meme conclusion a etc obtenue pour les femmes en ce qui conceme les composants passifs. Eur Respir J., 1989, 2, 522- 527.